Chat log from the meeting on 2016-11-22

From OpenSimulator

Jump to: navigation, search

[11:03] James.atLLOUD busy week - I see 42 items on the Git shortlog!
[11:05] Andrew.Hellershanks Nothing for 8 days then a burst of activity.
[11:06] Nebadon Izumi: ya I logged in a few minutes early so even if I got distracted id be here
[11:08] Andrew.Hellershanks Good plan
[11:08] Nebadon Izumi: Busy week here for me I have had a chance to get my server booting back up again to test opensim lately, i need to do that soon probably won't have much time until after OSCC though
[11:09] Nebadon Izumi: Things seem to be progressing well with OSCC though
[11:09] Andrew.Hellershanks I just received my welcome email with a link to the website that would allow me to set up my schedule.
[11:10] Nebadon Izumi: nice
[11:10] Nebadon Izumi: I need to spend a little time updating my displays in that grid
[11:11] Nebadon Izumi: maybe this weekend since its holiday week i should have a little time away from work
[11:11] James.atLLOUD heh - I have lot of items slated for these 'extra hours' - lol
[11:13] Basil Sosides: no Updates last week ?
[11:13] Nebadon Izumi: updates from what?
[11:14] Nebadon Izumi: opensim in general?
[11:14] Basil Sosides: 0.9.1
[11:14] Nebadon Izumi: oh yea I am not sure, Diva has not been around much lately
[11:14] Nebadon Izumi: i tried reaching her this morning with no luck
[11:14] Nebadon Izumi: so I am not sure what the status of the release is at the moment but hopefully soon'ish
[11:15] Basil Sosides: LAG on my Regions since weeks with the 0.9.1
[11:15] Nebadon Izumi: any clue maybe what is causing it?
[11:15] Basil Sosides: Internet is ok, under limit
[11:15] Nebadon Izumi: and have you tried master git as well?
[11:15] Nebadon Izumi: or the OSgrid release?
[11:16] Nebadon Izumi: just wondering if there is any difference
[11:16] Basil Sosides: i have that from the download site
[11:16] Alicia.Raven i had some serious LAG recently, but i found the cause and Ubit was able to fix, not sure if that fix is in the osg release
[11:16] Nebadon Izumi: I havent had a lot of time to test the latest code, I will try to soon though
[11:17] Basil Sosides: OSgrid (Dev) b6329fb: 2016-11-12
[11:17] Nebadon Izumi: ah ok that is good Alicia
[11:17] Nebadon Izumi: im sure Dan will pack one up soon enough
[11:17] Alicia.Raven UDP was slow because of blocking from some issues with user cache added last year
[11:17] Nebadon Izumi: ok cool
[11:17] Nebadon Izumi: well hopefully Basil new version helps
[11:17] Nebadon Izumi: keep an eye out for new OSgrid release
[11:17] Nebadon Izumi: you can use that in place of the 0.9.1 release
[11:17] Basil Sosides: ok, neb...thanks...i hope too
[11:17] Nebadon Izumi: to test if the lag goes away
[11:18] Robert Adams: what's "Basil"?
[11:18] Basil Sosides: LAG problems since weeks
[11:19] Andrew.Hellershanks Robert, Something used in cooking. ;)
[11:19] Robert Adams: ah... it's someone!! Hello Basil :)
[11:20] Basil Sosides: and some times happens NO Rebake when arrive my Sim
[11:21] Basil Sosides: Xbake dont start
[11:23] Robert Adams: baking has been a problem for a while... there have been timing problems before
[11:23] Basil Sosides: ah ok#
[11:24] Alicia.Raven i have arrived a couple of times as a grey avatar recently i think
[11:24] Basil Sosides: what about that UDP-Thing, is it fixed ?
[11:24] Nebadon Izumi: yea I get that as well even on 0.8.2
[11:24] Nebadon Izumi: i suspect it's the viewer itself
[11:24] Nebadon Izumi: the latest Singularity it happens almost always
[11:24] Nebadon Izumi: i try older singularity it's fine
[11:24] Andrew.Hellershanks Which UDP thing?
[11:24] Alicia.Raven yes, i found the commit out of testing past years worth of commits and Ubit found the problem
[11:25] Alicia.Raven its fixed in master
[11:25] Andrew.Hellershanks Alicia, What impacts did that UDP issue have?
[11:25] Basil Sosides: some UDP - Problems causes the LAG on Regions told some one
[11:25] Andrew.Hellershanks ok
[11:25] Alicia.Raven it was causing objects to populate VERY slowly when entering a region and generally lagging any UDP traffic
[11:26] Basil Sosides: yes
[11:26] Alicia.Raven the UDP packet queues emptied very slowly because of a blocking operation related to user account caching
[11:26] Andrew.Hellershanks I don't think I've noticed that but I'll do an update and check again to see what difference, if any, I notice.
[11:27] Alicia.Raven well Ubit almost didnt believe me as he said no one else had reported it, but we found it
[11:28] Basil Sosides: just try it on my Sim, 3 or 4 Avas then you'll see it
[11:28] Alicia.Raven the cause of it had been there for about 6 weeks tho, fixed it a couple of weeks back now
[11:28] Andrew.Hellershanks oh, ok. I've been using some older code so probably didn't run in to that problem.
[11:29] Alicia.Raven 6 months i mean
[11:29] Andrew.Hellershanks Oh, ok. I'm not that far behind.
[11:29] James.atLLOUD wild. could that 'blocking' work the other way - ie. UDP blocking user account cache... maybe also 'unknown user' symptoms?
[11:29] Andrew.Hellershanks I've been wanting to update a grid for a while. Now would be a good time to do so with that fix in place.
[11:29] Alicia.Raven i had to use git bisect to test a years worth of commits, then we couldnt understand how that commit could effect UDP the way it was
[11:30] Basil Sosides: the last functional Version without LAg was 57aac3d: 2016-08-14
[11:30] Alicia.Raven Ubit has fixed another problem with the unknown user names showing, i think name caps was asking the previous simulator after a teleport
[11:30] Andrew.Hellershanks Alicia: Does that mean you take over wearing the 'git bisect' crown from nebadon?
[11:31] Alicia.Raven well i didn't use git bisect, i just downloaded code and followed the same logic of try a version in between to see if it works lol
[11:31] James.atLLOUD <googles git bisect>
[11:31] Andrew.Hellershanks I saw mention of work regarding user names on the IRC channel.
[11:32] Andrew.Hellershanks Something about some data being cached and if there was a change the cached entry was being used instead of getting the proper (updated) information.
[11:32] Alicia.Raven yes, if you use the same viewer for avatars on different grids, the shared name cache can confuse knowing if people are local or HG
[11:32] Andrew.Hellershanks Alicia, ah, ok.
[11:33] Nebadon Izumi:
[11:33] Nebadon Izumi: for those wondering how to git bisect
[11:33] James.atLLOUD ty - very cool
[11:33] Nebadon Izumi: its a time consuming but useful way to find when things broke
[11:33] Alicia.Raven either the viewers or opensim has changed because radegast receiving IMs from HG users on their home grid now shows them as being local rather than showing a HG format name
[11:34] Nebadon Izumi: sometimes it can be hard to bisect, if you keep running into versions that do not compile successfuly
[11:34] James atLLOUD: OH! bisect is like finding a break in a cable! nice
[11:35] Andrew Hellershanks: A bit like doing a binary search of code.
[11:35] Andrew Hellershanks: James, it can be very handy to track down the source of an issue depending on how many code versions you have to check.
[11:36] James atLLOUD: ty
[11:36] Andrew Hellershanks: It takes time when you have to recompile between tests.
[11:36] Alicia.Raven only took 2-3 days to search a year of commits to find the cause
[11:36] Andrew Hellershanks: If the issue you are trying to reproduce doesn't isn't one that will be seen right away that just adds to the time it takes.
[11:37] Andrew Hellershanks: Impressive that you found it that quickly.
[11:38] Andrew Hellershanks: It would have been interesting to have stats from a load test before and after the change.
[11:38] Alicia.Raven lucky all i had to do was clear cache and enter, if it was an intermittent problem as you say it would take alot longer
[11:38] Andrew Hellershanks: Alicia, do you know how many code versions you had to check?
[11:39] Alicia.Raven umm no, 20 maybe
[11:39] Alicia.Raven rough guess
[11:39] Andrew Hellershanks: You did get lucky. Could have been a lot worse when you see how many code changes there have been in the past year.
[11:40] Alicia.Raven i started with 0.8 and moved to present day, each time testing a version in the middle of known working and broken ones
[11:40] Andrew Hellershanks nods
[11:41] Andrew Hellershanks: With Robert being here today I can put the spotlight on him once more.
[11:41] Alicia.Raven then when i found it, it was like "how can it be that commit?"
[11:41] Andrew Hellershanks: Robert, Anything new and exciting to report in the world of physics?
[11:42] James atLLOUD: Cool Alicia
[11:44] Robert Adams: nothing new... have been enjoying the holidays :)
[11:45] Andrew Hellershanks: Robert, are you working on BS these days (when you have time), or perhaps BS2, or working on both?
[11:46] James atLLOUD: December 10th approaches rapidly (OSCC day 1)
[11:46] James atLLOUD: <that was reply to Neb>
[11:48] Robert Adams: any work on physics has been on the existing BulletSim (fixing bugs) and on a totally different implementation using Bullet3
[11:48] Robert Adams: not very far along on the latter
[11:48] Andrew Hellershanks: ok.
[11:49] Sheera.Khan A question if I may... When I close a group chat window/tab in OS it reopens whenever there is a new message. In SL the group chat remains closed. It can be a bit distracting and annoying at times... Is that configurable or is it by design?
[11:50] Robert Adams: most of my coding time has been spent on a new asset system and mesh manipulation (auto prim merging and LOD genersion)
[11:50] Andrew Hellershanks: Sheera, that has been a long standing issue. I think it is a limitation of the group code. It doesn't seem to keep track that a person told the viewer to close the chat window.
[11:51] Sheera.Khan Thank you - too bad for me then ...
[11:51] Andrew Hellershanks: I have noticed it. It may be something I'll look at one day when I have time.
[11:52] James atLLOUD: I would have though that viewer specific.
[11:52] Andrew Hellershanks: I'm trying to remember if I saw reference to a related message about closure of a group chat window. Might have been something I saw in the group code last time I had looked at portions of it.
[11:52] Sheera.Khan it works differently here and in SL - using the exact same viewer ...
[11:53] James atLLOUD: huh, interesting.
[11:54] Andrew Hellershanks: James, in SL if you close a group chat window it won't open again when another message is posted to the group chat. You have to either open the group chat manually, or relog before it will open automatically once more.
[11:55] Sheera.Khan and given some drama queens and -kings that can be really helpful ^^
[11:55] James atLLOUD: agreed. havent' been in SL much (^_^)
[11:55] Andrew Hellershanks: yes, or when you are busy with something and don't want to be distracted by discussions in a group chat that is of little interest.
[11:56] Andrew Hellershanks: I may have a quick look at the group code after this meeting to see if I can find reference to a message that might be sent by the viewer when one closes a group chat window.
[11:57] Alicia.Raven osg doesnt use v2 groups does it?
[11:58] James atLLOUD: ah, I figured the viewer was ignoring subsequent notifications but the viewer actually tells the server to stop sending then?
[11:59] Andrew Hellershanks: James, I'm not 100% but I think that may be how it works.
[11:59] Andrew Hellershanks: or, how it should work if it was implemented.
[12:00] James atLLOUD: ty
[12:00] Andrew Hellershanks: Sheera, it is the type of change that should be isolated to a small section of code so it will be something I could handle. I don't know the deep innards of the code to tackle some of the big issues.
[12:00] James atLLOUD: ty Robert
[12:01] Andrew Hellershanks: WinGridProxy is a simple way to see if there is a group chat window close message.
[12:02] Sheera.Khan Cool that Andrew, that would be very helpful :-)
[12:02] Andrew Hellershanks: I'll add it to all the other things already on my todo list. :)
[12:03] James atLLOUD: oh, that list must be massive <sympathy>
[12:04] Andrew Hellershanks: James, My list is rather long as it includes items for another open source project and my own RL projects.
[12:04] James atLLOUD: Meeting attendance has dwindled. I hope my basic questions arn't too distracting.
[12:05] Andrew Hellershanks: I filled a sheet of paper with my todo list items once. It hasn't gotten all that much shorter.
[12:05] Andrew Hellershanks: np, James.
[12:06] Andrew Hellershanks: The people attending this meeting all have different levels of knowledge. Don't be afraid to ask something you think others will consider too basic a question.
[12:06] James atLLOUD: No, I understand. My work list keeps growing. I use Workflowy so I can filter and see things I meant to do months ago (gulp)

Personal tools
About This Wiki